search

tianzhi0549 / FCOS

FCOS: Fully Convolutional One-Stage Object Detection (ICCV'19)
https://arxiv.org/abs/1904.01355
Other
3.27k stars 630 forks source link

visual test results #150

Open 1356789-rr opened 5 years ago

1356789-rr commented 5 years ago

Hi! I've trained with my own data set.Now I want to know how to visualize the results of the test. I want to see the results of the test. In addition, I want to know how "thresholds_for_classes" infcos_demo. py are obtained, and how I can modify them if I want to use fcos_demo. py. thresholds_for_classes = [ 0.4923645853996277, 0.4928510785102844, 0.5040897727012634, 0.4912887513637543, 0.5016880631446838, 0.5278812646865845, 0.5351834893226624, 0.5003424882888794, 0.4955945909023285, 0.43564629554748535, 0.6089804172515869, 0.666087806224823, 0.5932040214538574, 0.48406165838241577, 0.4062422513961792, 0.5571075081825256, 0.5671307444572449, 0.5268378257751465, 0.5112953186035156, 0.4647842049598694, 0.5324517488479614, 0.5795850157737732, 0.5152440071105957, 0.5280804634094238, 0.4791383445262909, 0.5261335372924805, 0.4906163215637207, 0.523737907409668, 0.47027698159217834, 0.5103300213813782, 0.4645252823829651, 0.5384289026260376, 0.47796186804771423, 0.4403403103351593, 0.5101461410522461, 0.5535093545913696, 0.48472103476524353, 0.5006796717643738, 0.5485560894012451, 0.4863888621330261, 0.5061569809913635, 0.5235867500305176, 0.4745445251464844, 0.4652363359928131, 0.4162440598011017, 0.5252017974853516, 0.42710989713668823, 0.4550687372684479, 0.4943239390850067, 0.4810051918029785, 0.47629663348197937, 0.46629616618156433, 0.4662836790084839, 0.4854755401611328, 0.4156557023525238, 0.4763634502887726, 0.4724511504173279, 0.4915047585964203, 0.5006274580955505, 0.5124194622039795, 0.47004589438438416, 0.5374764204025269, 0.5876904129981995, 0.49395060539245605, 0.5102297067642212, 0.46571290493011475, 0.5164387822151184, 0.540651798248291, 0.5323763489723206, 0.5048757195472717, 0.5302401781082153, 0.48333442211151123, 0.5109739303588867, 0.4077408015727997, 0.5764586925506592, 0.5109297037124634, 0.4685552418231964, 0.5148998498916626, 0.4224434792995453, 0.4998510777950287 ]

tianzhi0549 commented 5 years ago

@1356789-rr these thresholds are printed after doing COCO evaluation on val set. If you cannot find these thresholds, 0.5 for all classes should be good.

1356789-rr commented 5 years ago

@tianzhi0549 if I train with my own data set and test with this demo, does the threshold need to change?

tianzhi0549 commented 5 years ago

@1356789-rr please try 0.5 first. If it cannot give you a satisfactory result, please tune it.

kyle1990kauffman commented 4 years ago

@tianzhi0549 where I can find the location of these thresholds in codes ?

Mahmood-Hussain commented 3 years ago

@tianzhi0549 where I can find the location of these thresholds in codes ? demo/fcos_demo.py

@tianzhi0549 I am getting the wrong labels for my images after training. My dataset contains different class names (labels) as compared to coco dataset.